Client- or Server-Based

The EIS design team must determine whether the EIS is client based or server based (see Figure 30.8). A client-based system will pull all data directly from the data warehouse and manipulate the data into usable form. The EIS would need to build the summary tables and store them inside the data warehouse or on the client itself. Queries issued by several clients will not cooperate and share resources in a client-based scenario.

Figure 30.8.  Client- versus server-based executive information systems.

For example, three clients may ask for the total sales in region 1. Each query performs the entire task of calculating the total sales at the detail level.

A better solution is to use a server-based EIS. The server-based EIS will be able to analyze the queries of all users and optimize those most frequently used. It will also be able to create summary tables, redundant tables, and other forms of denormalization at the server level. The EIS will be a completely different system than the data warehouse. This gives the EIS much more flexibility in restructuring data for speed and ease of use. The server-side EIS will cache query results from users. If one user requests the sales of toothbrushes in region 1, the EIS will calculate that result. If any other user requests the same data, the EIS will return the results quickly. This is especially useful when several departments are using the EIS. At various times in the fiscal year, multiple users issue the exact same query within days of each other. A query that may take several hours to run will only take that long for the first user who executes it. Subsequent users will receive results in seconds.

External Data

External data is pulled from various sources outside of the organization, which allows the users to see the overall business environment. For example, a graph that shows the sale of a particular product may have several confusing dips. However, when the external data is shown, it may be revealed that a competing product was heavily advertised during the periods of slower sales. Other external data includes the following:

•  Economic trends

•  Industry data purchased from information companies such as A.C. Nielson

•  Articles about the company or product

•  Competitor advertising

•  Competitor job advertisements

Data Mining (Panning for Gold)

Organizations have an enormous amount of data collected in the operational systems as well as in the data warehouse. Once the data is properly contained inside the data warehouse, it becomes easy for executives to ask data-related questions. However, executives do not know what questions they need to ask. Data mining searches though the data and finds relationships between different sets of data. If the relationships are considered valid, the tool alerts the executive team of its importance. Data mining is done by using complex statistical methods to determine relationships between the data. The relationships may be intra-table or inter-table. An easy relationship to see is the intra-table relationship between the city and the state inside the Customer table (see Figure 30.9).

Figure 30.9.  An intra-table relationship.

The data mining tool will show a high confidence factor between the relationship of a city equal to New York City and a state equal to New York. Since the confidence level is 100%, the user can assume that a relationship does exist. Unfortunately, the user has not learned much from this discovery. This relationship is well known and irrelevant to normal business.

The other type of data mining is inter-table. An example is the relationship between salespeople earning over $30,000 per year and the sale of product X. This type of data mining uses the join between the Salesperson table and the Sales table to determine a relationship (see Figure 30.10).

Figure 30.10.  An inter-table relationship.